Background: Rapid immunochromatographic tests (ICT) for dengue non-structural protein 1 (NS1) have shown good performance for diagnosing acute-phase dengue in serum in laboratory settings, but rarely have been assessed in whole blood and at point of care (POC). This study compare the accuracy and inter- and intra-observer reliability of the NS1 Bioeasy™ ICT in whole blood at POC versus serum in the laboratory, during a DENV-1 epidemic. Methods: Cross-sectional study involving 144 adults spontaneously demanding care in an emergency department within 4 days of onset of acute febrile illness. Accuracy of NS1 Bioeasy™ ICT was compared in whole blood and serum, both at 15 and 30 min, blinded to the reference RT-PCR or NS1 ELISA. Non-dengue patients were also tested for Zika virus with RT-PCR. Reliability of whole blood and serum readings by the same or different observers was measured by simple kappa (95% CI). Results: At 15 min, sensitivity (Sn) of NS1 Bioeasy™ ICT in whole blood/POC was 76.7% (95% CI: 68.0–84.1) and specificity (Sp) was 87.0% (95% CI: 66.4–97.2). Sn in serum/laboratory was 82% (95% CI: 74.1–88.6) and Sp 100% (95% CI: 85.8–100). Positive likelihood ratio was 5.9 (95% CI: 2.0–17.0) for whole blood/POC and 19.8 (95% CI: 2.9–135.1) for serum/laboratory. Reliability of matched readings of whole blood/POC and serum/laboratory by the same observer (k = 0.83, 95% CI: 0.74–0.93) or different observers (k = 0.81, 95% CI: 0.72–0.92) was almost perfect, with higher discordant levels in the absence of dengue. Results did not differ statistically at 5%. Conclusions: NS1 Bioeasy™ ICT in DENV-1 epidemics is a potentially confirmatory test. Invalid results at 15 min should be reread at 30 min. To optimize impact of implementing ICT in the management of false-negatives it should be incorporated into an algorithm according to setting and available specimen.